Gregory Hildstrom Projects Publications Resume Contact About Youtube Donate

Garage 1-Ton/12000BTU Ductless Mini-Split AC and Heat Pump Performance


I installed a Friedrich HM12YJA, 25 SEER 1-ton/12000BTU, ductless mini-split air conditioner and heat pump in my garage in July 2018. My friend Jimmy is also considering installing a garage AC unit, but we were both curious about performance numbers like recovery time. The 1-ton capacity I chose was a guess based on a performance estimate, a little over the 1 ton per 500 square feet rough guideline for south central Texas residential cooling. My garage is about 400 square feet, with a 9 foot ceiling, for about 3600 cubic feet total. It's pretty well insulated. Two walls and the ceiling are insulated and they connect to the house interior. The garage doors have R17 insulation.

Pre-Purchase Performance Estimation

I wanted a rough order of magnitude performance estimate before purchasing the unit. I wanted some clue as to whether recovery time would be measured in minutes or hours. Some searching led me to sensible heat calculation; there is another good discussion here.

Basically, in imperial units, BTU/hr = delta T * CFM * 1.08. Assuming adequate air circulation, we can solve for minutes, given BTU, delta T, and cubic feet: delta T * ft^3 * 1.08 / BTU = minutes. Or we can solve for delta T: minutes * BTU / (ft^3 * 1.08) = delta T.

I created the following estimation graphs for my garage and Jimmy's garage. The goal of this was to estimate the time required to cool 100F air down to a comfortable range. This does not take into account heat loss to walls, floor, or garage contents. This also does not account for latent heat of condensation. This is more of a rough recovery estimate, like after opening all garage doors for a few minutes and then closing them. That's enough time to exchange air, but not enough time to heat up the garage material. Zoom in or open the graphs in a new browser tab to see more detail.

I thought 10 minutes recovery time from 100F to 70F, at 12000 BTU, would be totally adequate for my purposes. Higher BTU seemed like a bit of overkill when I initially looked at these graphs. But again, these are air-only estimates in otherwise perfect, unrealistic, ideal conditions; so I knew I wouldn't hit these performance numbers.


Installation was relatively straightforward. There are a bunch of good videos on YouTube about installing ductless mini-split air conditioners. Instead of a plastic base, I poured a small concrete slab and I used concrete anchors for the outside condenser. For the inside wall-mount evaporator unit, I opted for left-side line exit and wall-mount lines. I wanted to take advantage of an existing vent hole in the bottom of the brick wall and I didn't want to place the wall unit in the corner. I also thought installation and maintenance would be a lot easier with wall-mount lines. Condensation drains just fine, but straight down from under the right side of the wall-mount unit would be ideal. I bought the system, line set, conduit, wiring, flare tool, vacuum pump, gauges, and disconnect box from AC Wholesalers. The pre-charge in the condenser was sufficient for my line distance, so I didn't need to add any refrigerant. The condenser is maybe 8' away on the other side of the wall. Condensation gravity drains to the other side of the wall with no pump.

Data Acquisition

I used a Raspberry Pi 3 and some DS18B20-based 1-wire temperature sensors. There's a good tutorial on using these two things together here. Here's my script:

for ((;;)); do

	DATE=`date +%Y%m%d%H%M%S`
	FILES=`find /sys/devices/w1_bus_master1 -name w1_slave`
	TEMPS=`cat $FILES | grep "t=" | sed 's/^.*t=//g'`

	echo -n "$DATE	"
	for T in $TEMPS; do
		echo -n "$T	"

	sleep 10

I pipe the script output to tee, so I can watch it on the screen and log it to a file. I converted the time stamps to elapsed seconds with some sed and calc post processing. I converted the raw data to degrees F in my gnuplot script. I used 10 temperature sensors in the same general chest-height area out of the air conditioner air flow and out of my circulating fan air flow. I used an Adafruit T-Cobbler for easier access to the R Pi GPIO pins.

Initial Cool Down, 99F ambient, Hot Garage, No Car

Ok, more BTUs would definitely help here, but I'm reasonably happy with this. I can take the preheated garage from high 90s to below 80, when ambient temperature is around 100, in one hour. Things are definitely comfortable by two hours.

10-Minute Doors-Open Recovery, 99F ambient, No Car

This test is closest to the pre-purchase estimation. We can clearly see the difference between ideal calculations and reality. Starting with a pre-cooled garage at 75, I opened both garage doors for 10 minutes. Ambient temperature was around 100, so air temperature rose fairly quickly. It peaked around 86-87. After closing the doors and turning the air conditioner back on, temperature was below 80 in 10 minutes, around 77 in 20 minutes, and back to 75 in 30 minutes. And the air is dry! This is totally acceptable. I'm happy, but again, more BTUs would recover faster. But we had to start this experiment somewhere.

Hot Car Recovery, 104F ambient

More BTUs for this torture test please. At 104 ambient, with a pre-cooled garage, I drove home from work, and I pulled my ridiculous hot Land Rover into the garage. I opened and closed one garage door as fast as possible to retain as much cool air as possible. Temperature didn't rise as high as I thought it would, but heat just kept pouring off of the vehicle. It was back below 80 in an hour and a half, down to 78 in two hours, and back to 75 in three hours.

BTU Output Measurement and Efficiency

I attached two k-type thermocouples to my ThermoWorks ThermaQ. The thermocouple have less mass than the rugged grilling probes that come with the ThermaQ. On a 96F day, I opened both doors and turned on surge mode. Using the sensible heat equation and the high fan speed CFM from the spec sheet, I was able to calculate BTU output. The spec sheet lists 598 CFM at high fan speed. I averaged several readings to get 88.35F average inlet temperature and 67.65F average outlet temperature. That's an average delta-T of 20.7F @ 598 CFM. BTU = CFM * delta-T * 1.08 = 598 * 20.7 * 1.08 = 13369 BTU. That's pretty close to the stated maximum output of 13500 BTU at unknown outside temperature, at 96F ambient. Right after doing the temperature measurements, I measured the current consumption at 4.2A. That makes for 13369 / (240 * 4.2) = 13.3 BTU/W efficiency at maximum output.